## Introduction

Nitrogen (N) is an essential element for plant growth, and its form and amount are mainly controlled by N production and consumption processes in natural soils1,2,3. In terrestrial ecosystems, if N cannot be efficiently conserved in soils, potential N losses will induce negative effects on the climate, the environment and even human health4,5,6. Therefore, it is essential to quantify the simultaneously occurring N production and consumption processes to identify whether soils can effectively conserve N.

Soil gross N transformations, driven by microorganisms, are greatly impacted by many soil properties and climatic conditions, such as soil temperature and moisture, soil pH, substrate concentration (NO3, NH4+ and organic N), and the quality and quantity of organic materials7,8,9,10,11. For instance, soil pH affects microbial community composition and activities, and is therefore important for NO3 production and consumption processes7,12,13. Soil carbon (C) and N availability can influence gross mineralization and NH4+ immobilization processes, which are important for assessing the indigenous soil N supply14,15. In general, the above mentioned factors are significantly affected by different land uses16,17,18, which thus induce differences in soil N transformations10,19,20.

With the development of 15N dilution technique and numerical model methods, studies on the gross N transformations in different land uses have been greatly increasing4,9,10,20,21. Generally, previous studies indicated that in natural forest ecosystems, most available N could be effectively conserved in soils through inherent soil N conservation mechanisms6,14. For instance, in subtropical forest soils, low gross mineralization rates combined with negligible gross nitrification rates were confirmed as an effective N conservation mechanism through reducing the production of inorganic N9. Furthermore, Zhang et al. found that the coexistence of high inorganic N production and immobilization rates was also an effective N conservation mechanism in subtropical forest soils4. However, after forestland conversion to cropland, the changes of vegetation cover and management practices (e.g. tillage and fertilization) could significantly influence soil N transformations16,17,19,20. In subtropical acid soils of Southwest China, Xu and Xu reported that compared to forestland soil, the cropland soils had significantly higher gross nitrification rates and lower immobilization rates, which resulted more NO3 losses through leaching or denitrification9. However, the mechanisms of different land uses influencing N transformation processes may be quite different according to soil types in subtropical regions. Therefore, more detailed studies are still needed to quantify the inherent N transformation processes under different land use soils in other important soil types in subtropical regions, which is beneficial for understanding whether those different soils can effectively conserve N.

The Sichuan Basin in Southwest China is a subtropical region characterized by numerous hills. Regosols (locally known as purple soil) are the most important and widely distributed cropland soils in this region, covering an area of more than 1.6 × 105 km2. Unlike the normally occurring soils in subtropical regions, Regosols are weakly developed mineral soils and are characterized by a shallow soil layer, coarse-textured sandy loams and good soil aeration22. Consequently, Regosols are susceptible to erosion and leaching, and the large N losses via leaching and overland runoff may cause local and widespread non-point source N pollution23. A previous study reported that annual nitrate leaching losses from sloping croplands could be up to 53.4 kg N ha−1 year−1 in this region, which was the dominant N loss pathway24. Mitigating this high loss rate needs to enhance the soil N retention capacity by understanding inherent soil N cycling mechanisms in Regosols. Many studies have indicated that soil N transformations predominantly regulate N forms and composition in soils14,25. In particular, Zhang et al. revealed that the NO3 proportion was mainly regulated by the process of nitrification26. However, the inherent soil N transformation processes are still not well understood in the Regosols of the Sichuan Basin. Wang et al. investigated the soil gross N transformations in cropland soils under different fertilization regimes in this region and found that the increased gross nitrification rates governed the increases in cumulative NO3 losses via interflow and overland runoff18. In the Sichuan Basin, the dominated land uses are forestland and sloping cropland. The difference in land uses can cause significant differences in soil properties27, which thus may affect soil gross N transformations. However, to date, how different land use soils influence gross N transformations in Regosols is not fully understood in this region. Understanding N transformations in Regosols and the effects of different land uses would provide important information for assessing the risks of N losses, and further provide the scientific basis for how to regulate the N transformation process to mitigate N losses in the Sichuan Basin of Southwest China.

In this study, we quantified gross N transformations in forestland and cropland of Regosols in the Sichuan Basin of Southwest China. This study aimed to (1) investigate the characteristics of gross N transformations in Regosols, (2) examine the differences in soil N transformations for different land uses in Regosols, and (3) evaluate the N conservation potential and N loss risks of Regosols under different land uses.

## Materials and methods

### Study region

This experiment was conducted at the Yanting Agro-Ecological Station of Purple Soil, Chinese Academy of Sciences, Yanting County of Sichuan Province (31° 16′ N, 105° 27′ E)23. It situates in the central Sichuan Basin and exhibits a moderate subtropical monsoon climate, with an average annual temperature and rainfall of 17.2 °C and 836 mm (30-year mean), respectively. The soil, classified as Eutric Regosols by the FAO soil classification28, is locally termed “purple soil” due to its purplish colour23. It is typically non-zonal and weakly developed mineral soil, mostly characterizing by a neutral or alkaline reaction18,23. Currently, forestland and sloping cropland are the main land uses in this area.

### Site description

To investigate the differences of inherent N transformations between different land use soils, sampling was conducted from the two main land uses of forestland and cropland. The forestland was initially planted with Alnus cremastogyne and Cupressus funebris in the 1970s to reforest cropland, and then, this site experienced natural succession without artificial management. It is now dominated by the representative forest type of Cupressus funebris in the study region, with a density of 1595 stems ha−1. The selected forestland site was with an area of approximately 1.3 ha−1. The cropland site was adjacent to the forestland site, with an area of 100 m × 100 m. It has been cultivated for more than 50 years, conventionally rotated with winter wheat (Triticum aestivum L.) and summer maize (Zea mays L.). Fertilizers in the winter wheat and summer maize seasons were applied at the same amounts of K (36 kg K2O ha−1) and P (90 kg P2O5 ha−1), but at different rates of N (130 and 150 kg N ha−1 as ammonium bicarbonate, respectively). All the fertilizers were manually applied and incorporated into the surface soil (0–20 cm) together with harrowing (approximately 20 cm deep). No irrigation was applied during either the wheat or maize season. The forestland and cropland sites had the same soil type (Regosols) and slope (5%).

### Field soil sampling and N loss monitoring

For forestland and cropland sites, grids with an area of 10 m × 10 m were uniformly divided. Then eleven and eight grids were randomly selected for soil sampling from different slope positions (i.e. upslope, middle slope, and downslope) in forestland and cropland sites, respectively, to minimize the potential effect of spatial heterogeneity on the experimental results. Furthermore, to minimize the effects of fertilization on gross N transformations in cropland soils, soil samples were collected in April 2016 when closed to the wheat harvest. At each sampling grid, the organic layer was removed first if present and three soil cores were randomly taken from the surface 0–20 cm layer. Then, they were well mixed, passed through a 2-mm sieve, and ultimately separated into two sub-samples. One sample was air-dried for soil property analyses, and the other sample was stored for the incubation experiment at 4 °C for < 1 week.

To better identify how soil N transformations regulate the mechanisms involved in NO3 losses, field monitoring of NO3 loss was conducted following each rainfall event during 2016. At each site, NO3 concentration in surface soil (0–20 cm) was also continually monitored at least once a week. Taking the selected sloping forestland site as a whole drainage area (1.3 ha−1), the discharge was monitored by the triangle weir installed at the outlet and the water samples were taken from the collecting tanks installed under the weir. At the cropland site, lysimeters were permanently established to take both interflow and overland runoff water samples23. Following each runoff event, the discharges of both interflow and overland runoff were determined. For both forestland and cropland sites, water samples were collected using 500-mL polyethylene bottles for assaying NO3 concentration. The annual NO3 loss flux (Q, kg N ha−1 year−1) was estimated as follow:

$$Q = \mathop \sum \limits_{i = 1}^{n} \left( {C_{i} \times q_{i} /100} \right)$$
(1)

where Ci is the NO3 concentration in the interflow and overland runoff sample (mg L−1), qi is the interflow and overland runoff discharges (mm), and n is the number of runoff events during the monitoring period. During the experimental period, the daily precipitation was automatically monitored by a meteorological station located at a distance of approximately 3 km from the sampling sites.

### 15N tracing experiment and model

The inherent gross N transformations in the forestland and cropland soils were determined using a 15N tracing technique. Two 15N sources, 15NH4NO3 (9.44 atom% excess) and NH415NO3 (9.75 atom% excess), were used in this study. For each soil sample (three replicates for each 15N labelling treatment), we prepared a series of conical flasks (250-mL) using 20 g of fresh soil (oven-dry basis) and added NH415NO3 or 15NH4NO3 solutions to obtain the same NH4+-N and NO3-N concentrations (20 mg N kg−1). Then they were adjusted to 60% WHC (water-holding capacity). In agricultural soils, the transformation of NH4+ to NO3 was generally fast6,21. To avoid the low NH4+ concentrations in soils after incubation and guarantee the 15N detection requirements29,30, therefore, relatively high application amounts of NH4NO3 were added compared to initial NH4+ and NO3 concentrations in this study. After 0.5, 12, 24, and 48 h of incubation at 25 °C, soils were extracted to measure the NH4+ and NO3 concentrations and isotopic composition. Distillation with MgO and Devarda’s alloy were conducted to separate NH4+ and NO3, strictly following the procedures described in previous studies31,32. Before separating, the recovery of NH4+ and NO3 in a standard solution (1 g NH4+-N/NO3-N L−1) was determined. The results showed that the recovery of NH4+-N and NO3-N in the solution was more than 99% and 95%, respectively. Finally, the 15N abundances of NH4+ and NO3 were analyzed by an automated C/N analyser and isotope ratio mass spectrometer (IRMS 20–22, SerCon, Crewe, UK).

The widely used numerical 15N tracing model was employed to investigate the gross N transformations in this study. For the details of this model, numerous previous studies can be referenced18,21,31,32. Briefly, this model mainly involved the following ten simultaneously-occurring processes (Fig. 1): MNlab and MNrec, mineralization of labile organic N and recalcitrant organic N to NH4+, respectively; INH4-Nlab and INH4-Nrec, immobilization of NH4+ to labile organic N and recalcitrant organic N, respectively; ANH4 and RNH4, adsorption and release of adsorbed NH4+ on cation exchange sites, respectively; ONH4, oxidation of NH4+ to NO3 (autotrophic nitrification); ONrec, oxidation of recalcitrant organic N to NO3 (heterotrophic nitrification); DNO3, dissimilatory NO3 reduction to NH4+ (DNRA); and INO3, immobilization of NO3 to recalcitrant organic N29,30. The transformation rates were calculated by zero-order, first-order or Michaelis–Menten kinetics by minimizing misfits between the modelled and determined concentrations and 15N enrichments of NH4+ and NO3 (averages ± standard deviations). Aikaike’s Information Criterion (AIC) was utilized to select the best model, and the Markov chain Monte Carlo-Metropolis algorithm (MCMC-MA) was employed for optimizing the parameter21,30,32. The MCMC-MA routine was conducted using MATLAB software (Version 7.2, The MathWorks Inc.). Finally, average transformation rates over a 48 h period (mg N kg−1 day−1) were calculated on the basis of the kinetic settings and the final parameters.

The total mineralization rates (MN) were estimated as the sum of MNlab and MNrec, and the total NH4+ immobilization rates (INH4) were calculated as the sum of INH4-Nlab and INH4-Nrec. Nitrification capacity was expressed as the ratio of ONH4 to MN. The NO3 retention capacity was defined as the ratio of NO3 consumption (INO3 + DNO3) to total nitrification (ONH4 + ONrec).

### Soil chemical property measurements

Soil property analyses strictly followed the procedures described in Soil Agro-Chemical Analysis33. Soil pH was measured in a 1:2.5 (soil-to-water) suspension using a DMP-2 mV/pH detector (Quark Ltd, Nanjing, China). Soil organic carbon (SOC) and total N (TN) were measured by wet digestion with H2SO4–K2Cr2O7, and by semi-micro Kjeldahl digestion using Se, CuSO4 and K2SO4 as catalysts, respectively. The concentrations of NH4+ and NO3 were measured by extraction with 2 M KCl, filtration with filter paper, and analysis using an AA3 continuous-flow analyser (Bran + Lubbe, Norderstedt, Germany).

### Statistical analyses

All statistical analyses were carried out in SPSS 20.0 software (SPSS Inc., Chicago, IL, USA). The Kolmogorov–Smirnov test was conducted first to test the normality of the data. The differences in the gross N transformation rates and soil properties between the cropland and forestland soils were detected using an independent sample t test. The significance level was conventionally set at 0.05. The relationships between gross N transformation rates and soil properties were tested by Pearson correlation analysis.

## Results

### Soil properties

The average pH was 7.85 ± 0.06 and 7.83 ± 0.03 in the forestland and cropland soils, respectively, with no significant difference (Fig. 2a). However, the average SOC content (22.91 ± 1.37 g kg−1), TN content (1.45 ± 0.13 g kg−1) and C/N ratio (16.26 ± 0.65) in the forestland soils were significantly greater than those in the cropland soils (mean 8.90 ± 1.21 g kg−1, 0.90 ± 0.07 g kg−1 and 9.72 ± 0.51 for SOC, TN and C/N, respectively, P < 0.05) (Fig. 2b–d). The average soil NH4+ concentrations were 2.46 ± 0.18 mg N kg−1 and 2.35 ± 0.07 mg N kg−1 in the cropland and forestland soils, respectively, with no significant difference (Fig. 2e). The average soil NO3 concentration in the cropland soils (13.60 ± 0.51 mg N kg−1) was significantly greater (P < 0.05) than that in the forestland soils (3.02 ± 0.08 mg N kg−1) (Fig. 2f).

### Soil gross N transformation rates

The gross N mineralization (MN) rates averaged 3.95 ± 0.51 mg N kg−1 day−1 and 4.84 ± 0.69 mg N kg−1 day−1 in the forestland and cropland soils, respectively, with no significant difference (Fig. 3a). No significant relationships were detected between the mineralization rates and the measured soil properties.

The gross NH4+ immobilization (INH4) rates in the cropland soils (0.11–0.92 mg N kg−1 day−1, mean 0.34 ± 0.10 mg N kg−1 day−1) were significantly lower than those in the forestland soils (3.56–10.31 mg N kg−1 day−1, mean 6.67 ± 0.74 mg N kg−1 day−1) (P < 0.05) (Fig. 3b). The soil gross NH4+ immobilization rates were positively related to the SOC content (r = 0.81) and the C/N ratio (r = 0.84) (Fig. 4).

Significantly higher gross autotrophic nitrification (ONH4) rates occurred in the cropland soils (7.91–20.02 mg N kg−1 day−1, mean 14.54 ± 1.66 mg N kg−1 day−1) than in the forestland soils (0.25–6.29 mg N kg−1 day−1, mean 1.99 ± 0.56 mg N kg−1 day−1) (P < 0.05) (Fig. 3d). However, gross heterotrophic nitrification (ONrec) rates were negligible in both the forestland and cropland soils (Fig. 3c). The soil gross autotrophic nitrification rates were negatively related to the SOC content (r = − 0.79) (Fig. 5a) and the C/N ratio (r = − 0.88) (Fig. 5b). Negative correlations also existed between soil gross NH4+ immobilization and autotrophic nitrification rates (r = − 0.75) (Fig. 5c). Nitrification capacity (i.e., ONH4/MN) in the forestland soils (mean 0.48 ± 0.12) was lower than that in the cropland soils (mean 3.94 ± 1.23) (P < 0.05) (Fig. 3g). A positive correlation was detected between the NO3 concentration and nitrification capacity (r = 0.89) (Fig. 5d).

The gross NO3 immobilization (INO3) and DNRA rates were not significantly different between the forestland and cropland soils (Fig. 3e,f). The cropland soils had a significantly lower NO3 retention capacity (mean value of 0.01 ± 0.01) than the forestland soils (mean value of 0.51 ± 0.24) in this study (Fig. 3h).

### Soil NO3− loss

During the whole year of 2016, six and seven runoff events were monitored in forestland and cropland sites, respectively (Fig. 6c,d). As shown, most NO3 losses mainly occurred in summer season (from July to August), during which the rainfall was usually heavy and contributed 51% of the annual precipitation (Fig. 6a). In particular, the heaviest rainfall event was observed at 2016/7/18 with precipitation of 162 mm (Fig. 6a). Then, the highest NO3 losses were monitored at 2016/7/18 in forestland and 2016/7/23 in cropland, respectively (Fig. 6c,d). The NO3 losses in each runoff event were all significantly higher in cropland than in forestland (P < 0.05). Significantly higher soil NO3 concentrations were also observed in cropland than that in forestland (P < 0.05), especially after the period of fertilization (Fig. 6b). The total NO3 losses were 0.25 ± 0.01 kg N ha−1 year−1 and 27.10 ± 2.54 kg N ha−1 year−1 for the forestland and cropland, respectively.

## Discussion

### Patterns of gross N transformations in Regosols

Different with the generally acidic and highly weathered soils in humid subtropical regions29, the studied Regosols inherit most properties of parent materials and characterize by coarse texture and a neutral or alkaline reaction18,22,23. The specific soil properties of Regosols therefore may cause different N transformation processes compared to the normally occurring subtropical acidic soils.

In current study, gross N mineralization rates in the forestland soils (mean 3.95 ± 0.51 mg N kg−1 day−1) were similar to those observed in subtropical zonal soils of Orthic Acrisols and Humic Planosols (FAO soil classification) (mean 3.67 and 3.52 mg N kg−1 day−1, respectively)4,34. But they were higher than those measured in subtropical acidic Regosols of southwest China (mean 1.23 mg N kg−1 day−1)35. This difference is most likely related to the differences in soil organic matter and pH between these two Regosols13,14. Moreover, the measured gross NH4+ immobilization rates in the forestland soils (mean 6.72 ± 0.74 mg N kg−1 day−1) were higher than those observed in above mentioned subtropical soil types (0.82 mg N kg−1 day−1 to 2.25 mg N kg−1 day−1)4,34,35. These results indicated that the alkaline Regosols might have a faster NH4+ mineralization-immobilization turnover than other soil types in the subtropical regions.

The average total nitrification rates in forestland soils were 2.00 ± 0.57 mg N kg−1 day−1 in this study, and autotrophic nitrification contributed approximately 99.6%. This result indicated that autotrophic nitrification was the dominated NO3 production process in the studied Regosols. However, significantly lower total nitrification rates were reported in subtropical acid soils, and heterotrophic nitrification was their dominant process4,34,35. Previous studies have showed that the microbiological autotrophic nitrification would be inhibited in soil at pH values lower than 536, but be stimulated at pH values higher than 7.537. Moreover, high soil pH might also inhibit the existence of fungi and their activities, which were related to heterotrophic nitrification7. Therefore, the differences in nitrification processes were likely related to the differences in soil pH between the alkaline Regosols and subtropical acid soils. Furthermore, the nitrification capacity (i.e., ONH4/MN ratio, mean 0.48 ± 0.12) in present forestland soils was significantly greater than that in acidic Regosols (0.02) and Orthic Acrisols (0.05)4,35, which may therefore promote soil NO3 accumulation and leaching risk in the study region22,23,24. This result could be verified by the high NO3/NH4+ ratio in forestland soils in this study.

Gross NO3 immobilization and DNRA were the important NO3 consumption and retaining processes in soils38,39. DNRA generally occurred in anaerobic conditions40,41,42, however, it was negligible in this study due to the good aeration of Regosols. Gross NO3 immobilization rates in this study were also significantly lower than those in other subtropical soils4,34. Previous studies have indicated that NO3 immobilization generally needed high carbon availability43. The forestland soils in this study had lower soil organic C content (22.91 ± 1.37 g kg−1) compared to those subtropical forest soils4,34, which thus likely resulted the lower NO3 immobilization. In addition, the inhibition of fungal activities by the high soil pH might also control the NO3 immobilization in the studied Regosols44. Consequently, NO3 retention capacity was significantly lower in alkaline Regosols (mean 0.51 ± 0.24) than that in subtropical zonal soils of Orthic Acrisols and Humic Planosols (0.98 and 0.81, respectively) under forestland4,34.

As discussed above, due to the specific soil properties, the non-zonal Regosols in the Sichuan Basin of Southwest China showed greatly different N transformation processes compared to the normally occurring soils in other subtropical regions. Overall, the alkaline Regosols had a faster NH4+ mineralization-immobilization turnover, higher nitrification rates, and lower NO3 immobilization rates compared to other reported subtropical soils.

### Gross N transformations under different land use soils

The different land uses significantly affected the NH4+ immobilization and autotrophic nitrification in Regosols, while no significant differences were observed in gross mineralization, NO3 immobilization and DNRA between the forestland and cropland soils (Fig. 3). In the cropland soils, the gross NH4+ immobilization rates were significantly lower than those in the forestland soils (Fig. 3b). This result agrees with most previous findings obtained from other subtropical soils in China9,45. However, Zhang et al. observed similar gross NH4+ immobilization rates between forestland and agricultural soils4. Soil gross NH4+ immobilization rates were positively related to the SOC and C/N in this study (Fig. 4). Previous studies indicated that relatively higher SOC content and C/N ratio in the soils could stimulate the increase of N immobilization potentiality9,46. In the cropland soils, due to long-term mineral N fertilizer application and very few crop residual retention, organic matter sources are mainly dependent on crop roots9. Therefore, the SOC content and C/N ratio in the cropland soils significantly decreased compared to those in the forestland soils (Fig. 2c,d), which might be an important factor that significantly reduced NH4+ immobilization. Furthermore, the gross NH4+ immobilization rates in the forestland soils were greater than the gross mineralization rates (Fig. 3a,b), indicating that a large proportion of the NH4+ produced from mineralization could be effectively immobilized. This rapid NH4+ turnover in forestland soils likely reduced the accumulation of NH4+, thus left little available substrate for nitrifiers19,25. However, in the cropland soils, the gross mineralization rates were significantly greater than the NH4+ immobilization rates (Fig. 3a,b), which might leave more available NH4+ substrates in soils for autotrophic nitrification.

The gross autotrophic nitrification rates were significantly greater in the cropland soils than in the forestland soils (Fig. 3d). This finding is in accordance with the results of some previous studies4,9. Soil pH is generally viewed as the key factor influencing nitrification4,12,13. However, soil pH was not significantly different between the forestland and cropland in this study (Fig. 2a). Nitrogen fertilization has been considered as another important factor stimulating nitrification in the cropland soils4,9. Numerous studies have indicated that long-term N fertilizer applications could stimulate autotrophic nitrification rates18,30,47,48. Applying mineral N fertilizer could induce the rapid increases in NH4+ concentrations for several weeks in cropland soils, thus providing sufficient available substrates for nitrification49,50,51. However, in this study, soil samples were collected once in April when several months had passed since the last fertilizer application. Thus, the extremely high nitrification rates following fertilization might not been considered in this study. Previous studies have shown that long-term N fertilizer application could affect the ammonia-oxidizing microbe population size and activity, thus stimulate autotrophic nitrification18,30. Previous studies in the same study area have revealed that application of mineral N fertilizer significantly increased soil ammonia-oxidizing bacteria (AOB) population size and changed AOB composition49,50. Consequently, long-term applying mineral N fertilizer might be the main factor inducing the differences in nitrification rates between the different land use soils in this study.

The ratio of soil gross autotrophic nitrification to gross NH4+ immobilization (ONH4/INH4) can effectively indicate their relative importance in NH4+ consumption5,9,10. In this study, a negative correlation was observed between the gross autotrophic nitrification rates and NH4+ immobilization rates (Fig. 5c). In the forestland soils, the average ONH4/INH4 ratio was 0.32 ± 0.09, indicating that NH4+ immobilization dominated NH4+ consumption. Conversely, the average ONH4/INH4 ratio was 69.99 ± 18.91 in the cropland soils, indicating that autotrophic nitrification was the dominant NH4+-consuming process. These results also implied that autotrophic nitrification was greatly enhanced in the cropland soils than in the forestland soils in the study region.

Overall, in comparison to the forestland soils, gross autotrophic nitrification rates were significantly increased, while gross NH4+ immobilization rates were significantly decreased in the cropland soils. The significant differences in soil N transformations were closely related to the long-term mineral N fertilizer application, and the significant SOC content and C/N ratio decreases in the cropland soils.

### NO3− loss and retention driven by soil N transformations

During the whole monitoring period, the cropland soils had significantly higher NO3 concentrations than the forestland soils (Fig. 6b). The large amounts of NO3 accumulation in the cropland soils could be easily diluted and lost during the heavy rainfall events22,23,24. The field monitoring results showed that the NO3 losses in each runoff event were all significantly greater in the cropland soils than those in forestland soils (Fig. 6c,d). Previous studies indicated that the inorganic N form and amount in soils, especially the NO3 accumulation, were controlled by N transformation processes25,26. In the cropland soils, the gross NH4+ immobilization significantly decreased but the gross autotrophic nitrification significantly increased compared to the forestland soils, resulting in the NO3/NH4+ ratio (mean 5.79 ± 0.16) being 4 times greater than that in the forestland soils (mean 1.24 ± 0.05). This result confirmed that NO3 not only was the dominant inorganic N form, but also has a higher concentration in the cropland soils.

The nitrification capacity (i.e., ONH4/MN ratio) and ONH4/INH4 ratio were two key indicators for the NO3 loss potential from the soils52,53. In this study, compared to the forestland soils, the cropland soils had much higher nitrification capacity and ONH4/INH4 ratio. The correlation analysis showed that the nitrification capacity and ONH4/INH4 ratio were positively correlated with the NO3 concentration and NO3/NH4+ ratio (P < 0.05; Fig. 5d). Moreover, the low NO3 retention capacity was observed in both the forestland and cropland soils (mean 0.51 ± 0.24 and 0.01 ± 0.01). Overall, the NO3 production rates were greater than the NO3 consumption rates, resulting in huge NO3 accumulation in Regosols (mean 3.02 ± 0.18 mg N kg−1 and 13.60 ± 0.51 mg N kg−1 for forestland and cropland), which thus caused great risks of NO3 losses, especially from the cropland soils9,18,54.

NO3 losses caused nutrient loss and threatened the environment and human health1,55,56. In this study, the NO3 losses occurring in the cropland soils were approximately 10% of the annual N fertilization. Consequently, NO3 losses to the environment should be minimized by retaining NO3 efficiently in the soils. As discussed above, the SOC content and C/N ratio significantly influenced soil N immobilization and nitrification. Thus, increasing the SOC content and C/N ratio would be effective strategies for NO3 retention in the alkaline Regosols, which can potentially reduce autotrophic nitrification and enhance NH4+ immobilization.

## Conclusions

Compared to the typical zonal acidic soils in the subtropical regions, the non-zonal soils of alkaline Regosols in this study showed specifically inherent gross N transformations, i.e. the faster NH4+ mineralization-immobilization turnover, the higher nitrification rates, and the lower NO3 immobilization rates. Different land use significantly affected the gross N transformation processes of autotrophic nitrification and NH4+ immobilization in Regosols. In the cropland soils, the rates of gross autotrophic nitrification were significantly greater, but the rates of gross NH4+ immobilization were significantly lower than those in the forestland soils. The specific soil gross N transformations resulted in low NO3 retention capacity and thus high NO3 loss risks in the Regosol croplands. The total NO3 losses from the cropland soils were substantial (27.10 ± 2.54 kg N ha−1 year−1) and much greater than those from the forestland soils (0.25 ± 0.01 kg N ha−1 year−1). The great differences in the N transformations between the different land use soils may be attributed to the changes of the SOC content and C/N ratio and the application of mineral N fertilizer after long-term cultivation in the cropland.